refactor(ai): remove OpenAI dependency, use only Ollama

- Remove OpenAI API key requirements and dependencies
- Make Ollama the default and only AI option
- Simplify workflow to use only local AI processing
- Remove OpenAI fallback logic
- Update documentation to reflect Ollama-only approach
- Clean up Python dependencies (remove openai package)

This eliminates external API dependencies and simplifies
the AI integration to use only local Ollama models.

.github/workflows/publish-live.yml

@@ -342,7 +342,7 @@ jobs:
           
           # Install Python dependencies
           python -m pip install --upgrade pip
-          python -m pip install openai requests
+          python -m pip install requests
           
           # Install Ollama
           echo "🤖 Installing Ollama..."
@@ -382,20 +382,9 @@ jobs:
             echo "✅ AI release notes generated successfully"
             cat "release_notes_${{ steps.version.outputs.new_version }}.md"
           else
-            echo "⚠️ AI release notes generation failed, trying fallback..."
-            
-            # Try without AI (basic analysis)
-            echo "🔄 Attempting basic release notes generation..."
-            python tools/scripts/maintenance/update_changelog.py \
-              --release-notes \
-              --version ${{ steps.version.outputs.new_version }} \
-              --previous-version ${{ steps.version.outputs.previous_version }} \
-              --description "${{ github.event.inputs.description }}" \
-              --openai \
-              --verbose || {
-              echo "⚠️ Basic generation also failed, using template"
-              # Fallback to basic release notes template
-              cat > "release_notes_${{ steps.version.outputs.new_version }}.md" << EOF
+            echo "⚠️ AI release notes generation failed, using template"
+            # Fallback to basic release notes template
+            cat > "release_notes_${{ steps.version.outputs.new_version }}.md" << EOF
           ## 📚 Release ${{ steps.version.outputs.new_version }}
           
           **${{ github.event.inputs.description }}**
@@ -423,7 +412,6 @@ jobs:
           ---
           *Note: AI analysis was unavailable, using template release notes*
           EOF
-            }
           fi
 
       - name: 📦 Create GitHub Release with PDF

docs/PUBLISH_LIVE_WORKFLOW.md

@@ -114,7 +114,7 @@ This analyzes:
 
 ### AI Model Configuration
 
-The workflow uses Ollama with configurable AI models:
+The workflow uses Ollama (local AI) with configurable models:
 
 - **Default**: `gemma2:9b` (fast, good quality)
 - **Alternative**: `gemma2:27b` (better quality, slower)
@@ -125,6 +125,8 @@ You can specify the model in the workflow inputs:
 - Set "AI model" field to your preferred model
 - Leave empty for default (`gemma2:9b`)
 
+**No API keys required** - all AI processing happens locally via Ollama.
+
 ### Manual Release Notes Workflow
 
 1. **Run publish-live workflow** → Creates draft release with AI notes

tools/scripts/maintenance/update_changelog.py

@@ -8,11 +8,8 @@ import requests
 import json
 from collections import defaultdict
 from datetime import datetime
-from openai import OpenAI
-
-# Initialize OpenAI client (will be None if not using OpenAI)
-client = None
-use_ollama = False  # Global flag to track which service to use
+# Initialize Ollama as default
+use_ollama = True  # Global flag to track which service to use
 
 CHANGELOG_FILE = "CHANGELOG.md"
 QUARTO_YML_FILE = "book/config/_quarto-pdf.yml"  # Default to PDF config which has chapters structure
@@ -752,7 +749,6 @@ if __name__ == "__main__":
     parser.add_argument("--demo", action="store_true", help="Generate a demo changelog entry with sample data.")
     parser.add_argument("-v", "--verbose", action="store_true", help="Verbose output.")
     parser.add_argument("-q", "--quarto-config", type=str, help="Path to quarto config file (default: book/config/_quarto-pdf.yml)")
-    parser.add_argument("--openai", action="store_true", help="Use OpenAI for summarization instead of Ollama (default).")
     parser.add_argument("-m", "--model", type=str, default="gemma2:9b", help="Ollama model to use (default: gemma2:9b). Popular options: gemma2:9b, gemma2:27b, llama3.1:8b, llama3.1:70b")
     parser.add_argument("--release-notes", action="store_true", help="Generate release notes instead of changelog entry.")
     parser.add_argument("--version", type=str, help="Version for release notes (required with --release-notes).")
@@ -805,10 +801,7 @@ if __name__ == "__main__":
         print("📝 CHANGELOG GENERATION CONFIG")
         print("=" * 60)
         print(f"🎯 Mode: {mode.upper()}")
-        if args.openai:
-            print("🤖 AI Model: OpenAI GPT")
-        else:
-            print(f"🤖 AI Model: {args.model} (via Ollama)")
+        print(f"🤖 AI Model: {args.model} (via Ollama)")
         print(f"🔧 Test Mode: {'ON' if args.test else 'OFF'}")
         print(f"📢 Verbose: {'ON' if args.verbose else 'OFF'}")
         print(f"📋 Features: Impact bars, importance sorting, specific summaries")
@@ -817,25 +810,17 @@ if __name__ == "__main__":
         
         print(f"🚀 Starting changelog generation in {mode} mode...")
 
-        if args.openai:
-            print("🤖 Using OpenAI for summarization.")
-            use_ollama = False
-            # Initialize OpenAI client
-            client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
-            if not os.getenv("OPENAI_API_KEY"):
-                raise ValueError("OPENAI_API_KEY not set. Please set it in your environment variables.")
-        else:
-            print(f"🤖 Using Ollama for summarization with model: {args.model}")
-            use_ollama = True
-            # Test Ollama connection
-            test_response = call_ollama("Hello", model=args.model, verbose=False)
-            if test_response is None:
-                print("❌ Failed to connect to Ollama. Make sure it's running on localhost:11434")
-                print("💡 To install models in Ollama:")
-                print("   ollama pull gemma2:9b")
-                print("   ollama pull gemma2:27b")
-                exit(1)
-            print("✅ Ollama connection successful")
+        print(f"🤖 Using Ollama for summarization with model: {args.model}")
+        use_ollama = True
+        # Test Ollama connection
+        test_response = call_ollama("Hello", model=args.model, verbose=False)
+        if test_response is None:
+            print("❌ Failed to connect to Ollama. Make sure it's running on localhost:11434")
+            print("💡 To install models in Ollama:")
+            print("   ollama pull gemma2:9b")
+            print("   ollama pull gemma2:27b")
+            exit(1)
+        print("✅ Ollama connection successful")
 
         if mode == "release_notes":
             # Generate release notes